AI and the Battle for the Soul with Stephen Wolfram and Iain McGilchrist - Conversation: What is Artificial Intelligence?

You know, Newton's mathematical principles of natural philosophy was, uh, sort of a surprise at the time.

You could use this idea of mathematics, you could predict things about the natural world.

People got used to the idea that science can predict things, that by using methods typically from mathematics, you could say, "We observe things, we now can predict what's gonna happen in, uh, in the motion of planets or the development of a pandemic or other kinds of things." You could...

That science would give one ways to predict things.

And that was an idea that, that, um, uh, worked in areas like physics, didn't work so well in areas like biology, where it's not really been possible to predict very much.

The, um...

And the question that I got interested in many years ago now is when we look at sort of complex things in the world, what will be predictable, what will not, and can one understand how to make a science that will talk about kinds of things which don't seem to follow the pattern of being, uh, predicted by mathematical equations? So one of the things I got interested in a long time ago is making theories for things.

What's the sort of foundation for that theory? And w- when you make a theory of something, you, you could imagine that there's some rule that the system, the, uh, the underlying system uses.

It applies that rule.

Perhaps it applies it over and over again.

That's sort of the essence of the idea of computation, is you're given a rule, you keep applying that rule.

So the question then is, when you do that, you might say, "Well, if I have a sufficiently simple rule, I'll be able to do the traditional science thing of jumping ahead and seeing what all the consequences of that rule are." But then you actually go and do the experiment.

I mean, I had the good fortune to be... live in a sort of time in history when one was first able to do these kinds of experiments.

It's kind of like instead of pointing the telescope out into the, into the, the heavens and seeing what's out there, you point a kind of computational telescope out into the computational universe and see what's out there.

And so I discovered a very surprising thing back in the 1980s, that, um, it...

when you might have thought you have a sufficiently simple rule, you would always have very simple behavior that would result.

But it isn't true.

And actually, we had a, we had maybe one, one picture that I might be able to show, maybe the first picture that we had picked out, if we're able to show it.

That's a good...

Yes, that's a convenient...

It's my all-time favorite science discovery.

Um, so the, that line of, uh, at the bottom is kind of a very simple rule.

You can think of it as a simple program that describes how the colors of cells will be determined, starting from the top, that one black cell at the top, and going down the page.

So that, that program, that rule at the bottom says how things should work.

That rule is very simple.